Weather whoopsie

Yesterday, Comrade Rowe was positively gushing during the Two-Hour Hate, remarking how astoundingly accurate the weather forecast was, 24 hours after it resumed coming from Gander, rather, he sneered, than from up in Halifax, Canada.

Today, the Ministry of Truth ran this somewhat more subdued, less exhuberant report:

Storm Peters Out
January 11, 2007

We could call it the storm that wasn't. The fact of the overnight snowfall was considerably less than the 20 to 25 cms forecast yesterday. Meteorologist Jason Shepperd at the Gander Weather Office indicated it was difficult to guage the track of the storm. He says about 10 to 15 centimetres fell on the Avalon for the most part, a little more in some areas. He says the storm tracked a little more east than expected. The gusting winds in the forecast also failed to materialize.

But of course Comrade Rowe could be foregiven for deluding himself, if only briefly, into thinking the quality of the forecast depended so heavily on the location where it was teletyped from.

After all, as the Prime Minister, the guy that Comrade Rowe and Glorious Leader shilled for in the last election, said last year:

For years now, this province has had to endure inaccurate weather forecasts from nearly a thousand kilometres away in the Maritimes... Particularly affected by inaccurate forecasts are the thousands of Newfoundlanders and Labradorians who work offshore... Newfoundlanders and Labradorians deserve better than to be told to expect five centimetres of snow over night only to wake up to ten times that amount.

The VOCM report sounds suspiciously like the media reports that were so frequent after the forecasting service was moved to Halifax in 2003. What the Ministry of Weather ordered, and what was delivered, were two different things. Anecdotally, some people thought that there was something bad about the Halifax forecasts, even in the face of meteorological reviews that found otherwise:

Environment Canada says its forecasts for two Newfoundland regions were on target this summer, despite a popular perception that its weather reports are often overcast with errors.

[...]

The Conservative government is the process of restoring public and marine forecasting at Gander, prompted by a petition that attracted 125,000 signatures, including Stephen Harper's, then the leader of the opposition.

The campaign picked up its greatest momentum when Environment Canada missed storms that dumped dozens of centimetres on snow on unsuspecting towns, or called for a blizzard that never materialized.

"I'd be surprised," said St. John's resident Jason Esdon, reacting to Environment Canada's accuracy statistics. Esdon feels that forecasts are correct only half the time.

The notion that the forecast is dependent on the location from which it is made is received truth.

Gander Mayor Claude Elliott said forecasting based in Gander will benefit the entire province.

"Safety is the biggest concern," Elliott said Tuesday. "I think we need that brought back so we can be more accurate because the weather changes so much in this province, that you have to be here to witness it."

And Harper himself bought that idea, if only for political expediency:

"We are deeply concerned about the safety of fishermen and all who work on the water or in the air," Harper said.

The Liberals cancelled forecasting in Gander. They moved marine forecasting to Halifax, 1,150 km away, and aviation forecasting to Montreal, 2,220 km away.

"Inconsistent forecasts and bad forecasts are more than a nuisance," Harper said. "They endanger the lives of people who work offshore, and that is unacceptable. The Liberals simply don’t understand that weather forecasting in this province is unique."

Never mind that Harper got the geography wrong. Or that agencies like the US Army and Navy have no problem generating forecasts for every continent and ocean from home soil. Halifax is just too far.

It is about 290 miles, as the crow flies, from the former met office in Halifax — Dartmouth, actually — to the closest point in Newfoundland, Cape Ray, near Port aux Basques.

And 290 miles, it turns out, is also the radius of the circle on this map:

That little red cross in the centre?

That's Gander.

If Halifax is too far from Newfoundland to accurately forecast the weather for any point in Newfoundland, even Cape Ray, does it not also mean that Gander is too far to forecast the weather for all but the southeastern extremity of Labrador? If proximity matters, then ask, should Halifax or Gander be issuing the forecast for Labrador City and Wabush, bearing in mind that the latter towns are geographically closer to the Dartmouth office than they are to Gander?

As the Prime Minister said last year in his "Newfoundlanders and Labradorians deserve better" speech:

What is relatively new though is scientifically accurate weather forecasting technology — a technology our government believes the people of Newfoundland and Labrador should be able to benefit from.

Weather forecasting has always been, and likely always will be, a blend of art and science. But as new technologies are deployed on the ground, on the oceans, in the atmosphere, and above it, and as new computing hardware and methods are developed to crunch and model the data, it's increasingly a science.

Anton Chekhov said, "There is no national science, just as there is no national multiplication table; what is national is no longer science."

When it comes to the science of meteorology, Gander is no worse a location than Halifax to forecast the weather in Nain; Gander residents are no worse at doing it than Montrealers. But, setting aside the politics, scientifically, it is no better. The Gander weather office could just as easily have been relocated to Stephenville or St. Anthony or Wabush.

However distinctive or fickle a place's weather, there is no national meteorology.

It would be interesting to set aside the type of anecdote that Bill Rowe, Jason Esdon, and others, engaged in from 2003 to 2007, and carry out an impartial, scientific, mathematical analysis of past weather forecasts for a representative sample, geographically and chronologically, of Newfoundland and Labrador weather. The ones made by Gander up to and including 2003. The ones made in Dartmouth from 2003 to 2007.

And then, to compare them to Gander, 2007 and beyond.

If the location of the met office is such a critical factor in weather forecasting accuracy, it will show up in the numbers when you compare Dartmouth with the old Gander records.

It should also, this being science, show up in a reduced accuracy the farther afield the forecast area under examination — say Nain to Hopedale, or Wabush and Vicinity — is from the office forecasting the weather. This is, after all, the working hypothesis: that when it comes to forecast accuracy, proximity matters.

Was the Gander issue about politics and economics? Or about the science of meteorology? The numbers wouldn't lie.

It's the sort of study that should be right up the alley of The Leslie Harris Centre of Regional Policy and Development.