The long, hot summer

RUTH.BONNEVILLE@FREEPRESS.MB.CA Enlarge Image

A young couple watches the sunset at Grand Beach on Lake Winnipeg as the summer of 2009 comes to a close.

The weather record on Winnipeg's endless summer makes it pretty clear how good we've had it from June through to the end of August.

All told, we had 19 days where the daytime high got above 30 C. Normally we only get 12. And how about this measuring stick? Our 19 days of plus-30 C matches the total for the past three summers combined.

And when you add it up, our temperatures were 1.3 C warmer than average, overall.

Better still, a summer that brought 60 per cent less rain than normal gave us some splendid weekends. There was rain on only four Saturdays or Sundays out of 26 since June 1, says senior Environment Canada climatologist Dave Phillips. "Canadians live for the summer weekends."

The hottest day this summer -- and the hottest in 16 years -- was Aug. 23 at 37.2 C.

But not only was this summer serving up the right amount of heat and sun, it turns out Environment Canada was pretty much right, at least most of the time when it came to forecasting the temperature.

The Free Press compared six-day weather forecasts against what the actual highs and lows ended up being for 67 days between June 15 and Aug. 21. On average, forecast highs and lows were correct to within 2.5 C. One-day forecasts were correct to within 2.1 C, while six-day forecasts were accurate to within 2.9 C.

Those numbers are actually better than what Environment Canada aims for -- accuracy to within three degrees.

For a comparison, we looked at six-day forecasts for 45 days in the same period from the U.S. National Weather Service for Grand Forks, N.D. On average, the U.S. temperature predictions were correct to within 2.2 C -- about 0.3 C more accurate than the Canadian ones.

National Weather Service meteorologist Gregory Gust says the Canadian results are actually very good considering the amount of territory the different stations must cover. The Grand Forks station serves an area around 115,000 square kilometres in size (close to the size of Pennsylvania), while the Winnipeg and Edmonton forecast centres together cover the entire Canadian Prairies and Arctic.

The population density in North Dakota isn't high, but weather doesn't start where it ends, and the need for accurate predictions on the U.S. eastern seaboard funds a high density of meteorologists across the entire United States. For instance, to help with hurricane modelling during the recent preparation for hurricane Irene, Grand Forks was flying six weather balloons instead of two, Gust says.

But Environment Canada meteorologist Ken Macdonald doesn't agree with Gust's theory, and says instead the U.S. forecasts could be more accurate because they are issued in degrees Fahrenheit, which is a finer measurement.

While Environment Canada was good at predicting temperatures overall, that's not to say it wasn't completely wrong occasionally. For example, for Aug. 16, it predicted a high of 21 C one day earlier, and then the temperature rose to 27.5 C. For June 28, the forecast six days previous had called for a low of 16 C, but the mercury ended up dropping to 3.8 C -- meaning they missed the mark by more than 12 C.

Beyond temperature, predictions of precipitation and weather conditions were harder to measure, since forecasts deal in percentage chance of precipitation, and measuring sky conditions is subjective. But the Free Press looked at how many times the conditions forecast were completely wrong for 49 days in June, July and August. For instance, occasions where the forecasts predicted sunny skies, and in fact, it rained or was overcast, or where a prediction of rain or overcast skies saw mainly sunny ones.

We found Environment Canada was extremely accurate in the short term, but that things got a little dicey as it became more long-term. One-day forecasts were completely wrong only 12.2 per cent of the time, on average, but six-day forecasts were wrong 44.9 per cent of the time.

There was the occasional day where the forecast was wrong for many days leading up to it. For July 17 and 18, Environment Canada predicted completely sunny skies for both days for all six days leading up to them, only for it to be mostly cloudy in the middle of the day, both days. At other times, the agency was bang on: it predicted sunny skies for Aug. 2 during the six days leading up to it, and it was indeed sunny that day at noon.

Environment Canada doesn't record the accuracy of its forecast conditions, but in terms of precipitation, its forecasts are more accurate than what the climatology (the historical norm for a certain day) would predict, nine times out of 10, Macdonald said.

william.burr@freepress.mb.ca daniela.germano@freepress.mb.ca meghan.potkins@freepress.mb.ca

Average accuracy of forecast

As we get further from the actual forecast date, both Environment Canada and the National Weather Service see the accuracy of their forecasts decrease.

Predicted vs. actual conditions

This chart shows the percentage of days Environment Canada's forecasted conditions differed significantly from the actual conditions on a day at noon -- for example, when it is rainy and the forecast called for sun, or vice-versa

Republished from the Winnipeg Free Press print edition September 1, 2011 A6