Fall 2015: How our forecast compares with the competition
Visit this Fall Forecast Guide to the Season for the Fall Forecast, Winter Weather Preview and more.
Thursday, October 8, 2015, 9:34 AM - The Weather Network Forecast for Fall 2015 reveals extended warmth for some and a possible 'false start' to winter later in the season. How does our forecast compare with the rest?
What everyone is saying
Check out the full, in-depth story on The Weather Network's Fall 2015 Forecast, including the regional breakdown, however here is the basic pattern expected for the months to come:
As is typical in the fall, the basic pattern of temperatures is an average, but different parts of the season will see different trends, as revealed in the breakdown between "early" Fall and "mid-to-late" Fall, below:
"Fall is a transition season and we typically see a roller-coaster ride in temperatures as we descend toward winter. This year, Canadians should expect more prolonged periods than usual in which temperatures are either above or below normal," said Chris Scott, chief meteorologist at The Weather Network. "Above-seasonal temperatures will dominate central and eastern parts of the country in September, but a switch will flip sometime in October, bringing cold temperatures and a false start to winter. By contrast, Western Canada will hold onto milder weather longer into the fall, resulting in above-normal temperatures for the season as a whole."
Environment Canada's Sept-Nov Seasonal Forecast
Temperature and precipitation forecasts from Environment Canada (shown above and below, respectively) have two maps each. The "deterministic" forecast (left) predicts the expected conditions - above, below or near normal. The "probabilistic" forecast (right) shows the predicted likelihood that those conditions will occur, thus giving the forecast confidence level.
AccuWeather's 2015 Fall Forecast
How each forecast is made
Some forecasts are produced using a very hands-on approach - analyzing times in the past when there were similar patterns in weather, teleconnections and sea-surface temperatures, and then taking into account how current differences from those past conditions will affect the outcome. Some are the result of complex sets of computer models, generating an "ensemble" look at the future weather, which emphasizes the similarities between the various model results. Others take a combined approach that considers both analogue years and computer model results.
According to AccuWeather's corporate website: "Many industry leaders consider AccuWeather seasonal forecasts to be their best-kept strategic secret." Thus, the company declined to discuss their forecasting methodology.
In the past, Environment Canada relied on its forecasters taking a very hands-on approach to their seasonal forecasts. Computer model results were adjusted based on their input, typically after examining analogue years for similar weather patterns. However, the decision was made roughly 15 years ago to switch to a pure computer modelling approach.
According to the Environment Canada website, seasonal forecasts are the result of 20 different model runs — 10 each from two different coupled atmosphere-ocean models (CanCM3 and CanCM4).
Forecasters meet to discuss the results before they are released to the public, according to Environment Canada climatologist David Phillips, however no changes are made to those results.
The reason for this, Phillips explained, is to remove the human influence from the forecasts, in order to better evaluate the performance of the computer models.
However, Environment Canada's forecast models are run on a daily basis, and their seasonal forecast website is updated every month. Thus, if Phillips is called sometime in the middle of the season for a specific forecast, he will not only be drawing on his vast knowledge of Canadian weather history, but he will have access to the latest model runs as well.
The Weather Network
The Weather Network's forecasting team, led by Dr. Doug Gillham, produces seasonal forecasts using a balanced approach. The team analyzes global weather and jet stream patterns, as well as ocean water temperatures and what are known as "analogue years" - years in the historic weather record that best match the current patterns.
"While no two years are identical, we can learn a lot about upcoming seasons by looking back to see what similar patterns have produced in the past," Gillham says. "The biggest challenge is in correctly identifying the key factor that will drive the weather pattern during the upcoming season."
Our team also considers computer model guidance that is produced by weather services around the world. These computer models assist us in determining to what extent the current pattern will persist versus how it will change during the weeks and months ahead.
"Understanding the current pattern and how it should evolve is critical to being successful, as the models will often disagree with each other or come up with a forecast that does not make sense meteorologically," Gillham added.
Forecasting for a "transitional season" - Spring or Fall - is inherently more difficult than for Summer and Winter, however let's take a look at the temperature forecasts for Summer 2015.
There was a general consensus between The Weather Network, Environment Canada and Accuweather that it was going to be cooler than normal in the northeast of Canada and warmer than normal in the west. However, the forecasts differed quite a bit for northwestern, central and eastern Canada.
How did everyone do?
The above forecasts are fairly meaningless unless you look at what actually transpired during the season.
Looking only at temperatures for this comparison, here is what was recorded for the months of June, July and August, across North America:
Credit: WeatherBell, with edits by author
It will be a few more days, at the very least, before a seasonal map - compiling all of June-July-August temperatures - will be available for a more direct comparison. However, looking at all three months together, above, it would appear that the consensus between the three forecasts was generally correct: It was generally warmer than normal in the west, and cooler than normal in the northeast.
In Atlantic Canada, the season started off quite cold, compared to normal, and especially so for Newfoundland and eastern Labrador in July.
In Ontario and Quebec, though, this region showed the most profound difference between the forecasts. The wide swath of "cooler than normal" temperatures in The Weather Network's summer forecast - stretching from the US southwest, through the midwest, over the Great Lakes and through Ontario and Quebec - matches up with the observed pattern.
Why do we do this?
Overall, the goal of seasonal forecasting is to effectively describe the dominant weather patterns that are expected over the next three months. This year has already posed significant challenges with seasonal forecasts and will likely continue to do so.
The "Blob" of warmer ocean temperatures off the Pacific coast is one reason for this. Another is the very unusual El Niño in the equatorial Pacific Ocean - which not only developed extremely late, mimicking the strange '82-'83 El Niño, but which continues to push forecasts towards record-breaking levels.
These factors are testing computer models to their limits, and making those forecasts with a human touch far more accurate.
Note: While a forecast from the Farmer's Almanac was included in our Summer Forecast, the Farmer's Almanac does not provide a dedicated Fall forecast for comparison in this piece.